AI-driven search is a big part of those challenges. As of October 2025, a full third of the 15 billion searches conducted on Google (or 5 million) triggered AI Overviews, Semrush president Eugene Levin told Digiday. At the same time, a significant percentage of publishers have said their traffic is decreasing.

Well over half of publisher respondents to Digiday+ Research’s fourth-quarter 2025 survey said they saw traffic declines last year: 54% said their traffic decreased somewhat or significantly throughout the year in 2025.

In fact, traffic pattern changes specifically related to social media and AI and site traffic declines due to AI were the top two challenges to the publishing industry identified by respondents to Digiday’s Q4 survey: 25% said traffic changes were the biggest challenge and 16% said AI-driven traffic declines were the biggest challenge.

And that trend is likely to continue into 2026. Nineteen percent of survey respondents each said traffic changes due to social media and AI and AI-driven traffic decreases will be the biggest challenges for the publishing industry this year — marking the largest groups of respondents in Digiday’s survey.

In this second part of our report on AI’s effects on the media industry, Digiday explores how publishers are navigating the rapid changes to search as AI-driven platforms reshape how people access information and how publishers monetize content.

Digiday+ Research surveyed 40 publisher professionals in the fourth quarter of 2025 about their use of AI and current and future investments in the technology. Digiday+ Research also conducted individual interviews with publisher executives responsible for AI investments and applications development. They are executives from:

• Business Insider
• Dow Jones
• Forbes
• People Inc.
• Really Simple Licensing (RSL)

The shift to AI-powered search has been fraught with challenges for publishers. According to a report from analytics and licensing platform Tollbit, AI chatbots drive significantly less traffic to publishers than traditional Google search — 95.7% less on average as of Q4 2024.

International trade association Digital Content Next, which counts the New York Times, Condé Nast and Vox among its approximately 40 member companies, also found that publishers’ organic search referral traffic from Google has declined broadly. The majority of DCN member sites experienced traffic losses from Google search between 1% and 25% as of the summer of 2025. 

AI also affects publishers’ ability to control content scraping. A September 2025 report from Tollbit found that 1 out of every 50 website visitors could be attributed to AI in Q2 2025, up from 1 out of every 200 visitors in Q1 2025. Over the same period, human site traffic decreased by 9.4%.

“This year, organic traffic has declined sharply as generative AI tools increasingly intercept the path to publisher content,” Nada Arnot, evp of marketing at The Economist, told Digiday in December 2025. “The promised upside — traffic from LLMs and AI-powered search — hasn’t materialized. For publishers who rely on discoverability, that’s a growing concern.” 

“If AI is to be part of a sustainable media ecosystem, it must also support the sources that fuel its value,” she added. 

In response to the growth of AI-powered search, Digiday asked publisher respondents how they are approaching generative engine optimization and answer engine optimization — which are, essentially, to AI-generated search what SEO is to traditional search. 

More than one-third of publisher respondents (40%) said that their companies have not changed strategies to optimize for GEO and AEO, while 13% of respondents said that they don’t know what GEO and AEO are. 

In total, slightly more than half of publisher respondents (53%) said their company is not optimizing for GEO and AEO, despite the decline of traditional search as a traffic referral source — a high percentage given the extent to which publishers are being affected by AI search.

Among publishers whose companies are taking action to optimize for GEO and AEO, the top three strategies publishers said their companies are using are conversational style formats like Q&A (30% of respondents), structuring site metadata for better readability by machines (27% of respondents) and increasing use of subject-matter terms (23% of respondents). 

Nina Gould, chief innovation officer at Forbes, said an article’s format tends to drive the publisher’s GEO strategy. “We see resonance in areas where our brand has depth like investing, but we also see that when we help the LLMs match to a prompt and then define the answer, that content gets cited a lot and tends to drive clicks,” Gould said. “It is, ironically, not that dissimilar to SEO practices, where you want to have a clear heading defining what the next section of an article is referencing. That seems to help guide LLMs, because they can match the prompt that a user has put into that heading. Then it knows clearly what section of the article is answering that prompt.”

However, even when publishers optimize for GEO, they struggle to measure the effectiveness of their strategies. Unlike with traditional SEO, AI platforms and technology companies don’t share AI search data directly with publishers. While there are companies that track and report on AI search, similar to Semrush and Similarweb for SEO, many of them are not able to provide 100% accurate analysis.

“Every data provider is in the same boat. We just don’t have perfect information right now,” Lily Ray, vp of SEO strategy and research at performance marketing agency Amsive, told Digiday in October 2025. “I think that’s one of the biggest challenges. Even if you invest a ton of money into AI search, your reporting is always going to be questionable right now. The methodology is always going to be questionable.”

Adelle Kehoe, director of product marketing at Similarweb, said that, while AI search data is largely directional, it’s as good as it can be for the time being. “[AI] answers are probabilistic, so this is very directional at the moment,” Kehoe said. “We’re not saying for every person in the U.S., they’re going to receive the exact same answer, but it is how businesses can get the best idea of how they’re ranking or if they’re in the conversation or not.”

Publishers are also finding that AI-powered search does not drive conversions. According to SEO marketing company BrightEdge, which looked at data from January 2025 to August 2025, AI search accounted for less than 0.1% of traffic among the thousands of queries and websites analyzed. Organic search was the primary driver of conversions during that period, according to BrightEdge.

Some publishers have made moves to partner directly with AI platforms in the form of content licensing agreements in which publishers let AI companies use their content to train LLMs, often including paywalled content. In exchange, publishers get attribution for content surfaced by the AI chatbots or search platforms, as well as access to technology that publishers can use to build AI-powered products and features. 

Harry Hope, CTO at Business Insider, said the publisher has concerns about how AI search is evolving, but that Business Insider’s parent company, Axel Springer, has entered into a partnership with OpenAI. 

“We want to support the work that our newsroom does, and to pay everyone, and so we believe that our content needs to be fairly compensated,” Hope said. “If these various tools, LLMs and search engines want to consume our content, we’re optimistic that there’s a path to a fair marketplace, and a way that our content can be compensated appropriately and fairly.”

Licensing agreements between publishers and AI companies vary widely. For example, publishers including Business Insider and People Inc. signed licensing deals with Microsoft in 2025 to participate in the tech company’s “Publisher Content Marketplace” pilot program. Through the program, Microsoft is creating a pay-per-usage AI content marketplace to compensate publishers when AI-backed products use their content. 

People Inc. CEO Neil Vogel described the Microsoft marketplace during parent company IAC’s third-quarter earnings call last year as an “a la carte” pay-per-use model, in contrast to the “all you can eat” lump sum deal People Inc. has with OpenAI. “We are very happy with either model — both can be viable as long as our content is respected and paid for,” Vogel said.

People Inc. Chief Innovation Officer Jonathan Roberts said the publisher closely monitors its partnership with OpenAI and the platform’s access to its content. “We track it monthly, and we talk to them weekly about where this is going, because we’re at the beginning of an innovation in AI search, not at the end,” Roberts said. “I’m very happy to see that AI search has decided across the board that they should cite their sources.” 

On other hand, Roberts said other AI companies continue to scrape People Inc.’s content without permission. “They use it for things that we don’t allow, and they claim that they have free use to do this,” Roberts said. “That’s like if the Yellow Pages were to list your store, and then the Yellow Pages steals items off your shelves and sells it outside your front door.” 

“As we move into 2026, publishers along with AI companies need to rebuild a new, sustainable information economy,” Roberts added.  

For its part, Really Simple Licensing is working to create standards for publisher compensation when it comes to these content licensing agreements. 

RSL designed its licensing framework to standardize how publishers tell AI systems what content they can use, and how publishers should be compensated. “The RSL standard is a machine readable licensing standard so that anyone can describe the license terms for their content in a way that machines and crawlers can access and comply [with],” RSL founder Doug Leeds explained. 

“That infrastructure layer is something that’s been missing from the web,” he added. “AI abstracts take publishers’ content and they’re not getting traffic back. They can’t monetize it, and the ecosystem is falling apart. So how do you articulate your license terms? We built RSL as a standard to do that. It’s open, it’s free.” 

Leeds also co-founded RSL Collective, a nonprofit collective rights organization that centralizes how publishers, creators and AI companies enter content licensing agreements. “If I want all the content on the internet, does that mean I have to sign a contract with everyone?” Leeds asked. “RSL as a collective management organization is offering a solution by which you [AI companies] can license all the content on the internet with one deal. It takes away the cost of doing those deals from both sides, and allows you [AI companies] to have actual licensed content that supports the creators of the content that you’re using for your product.”

Attribution is another challenge for publishers in the era of AI-powered search. Currently, there are no incentives for tech companies to share visibility, traffic or conversion data with publishers. 

According to Leeds, it would be mutually beneficial for AI companies to share that information with publishers. “If we can efficiently give money to content sources that are being used in AI, it’s in [the AI company’s] interest to describe who those are,” Leeds said. “Right now, with no licensing, it’s not in their interest because they’re basically telling you [a publisher], ‘We took your content and we didn’t pay for it.’ But once it’s being paid for, then attribution becomes in their interest, because they say, ‘OK, these are our inputs. Please make sure that you continue because we’re using them and they’re driving value.’”